The larger content size is due to LOD, multiple textures, and other stuff. This is the same with all other content.
The trees themselves aren't really on the route. They are referenced in the map.obs file (objects file), but exist in their own content folders elsewhere in your Trainz content on disk.
The trees, or any object for that matter, gets loaded into memory and graphics memory. Once loaded there should be little impact in TRS2019. This wasn't always the case with TS12 and below. Since content is cached by the system, if you use more than one of the same tree, such as in a pine forest, you'll have only one "real" tree loaded and the others are referenced in memory off of that tree. These other trees are called instance objects in the 3d modeling and animation world.
Back in another life way back in 1995, I worked for a training company that specialized in plastics manufacturing. I worked with the 3d modeler and we built the animated scenes and training program products. There was one particular introduction showing a room full of highly detailed injection molding machines. In that scene was only one "real" model. All the others were instance objects. If the modeler had loaded up a room full of actual models, the computer couldn't render the animation due to the amount of processing power needed to animate each and every machine. Only one machine was animated physically using the same kinds of things we do for locomotives and other animated assets. The instance objects followed along and performed the same animation at the same time because they were linked to the main object.
Keeping that in mind. You'll find you'll have better performance using a single tree or a small number of trees for the forest rather than a huge amount of different trees. This holds true for towns, cities, and other places where there's lots of repetitive objects. When building a town, I will put various detailed objects near the tracks, but in those locations away from the tracks, I will use the same object to fill in. This reduces memory overall.
Hope this explains it.